Keeping a clear head when everything accelerates: what your brain cannot do

A leader losing clarity does not stop deciding. That is the whole problem.

They keep making calls at the same rate, answering as fast, holding their meetings. Nothing collapses. What changes shows up three weeks later, in the number of reopened decisions, in projects relaunched twice, in that diffuse sense that the company is turning without advancing.

Visible performance holds. Accuracy gives way first.

Interruption does not slow you down, it makes you wrong

The common belief is that acceleration costs time. The data says otherwise, and it is not new.

In 1999, Cheri Speier, Joseph Valacich and Iris Vessey published in Decision Sciences an experiment distinguishing two kinds of task. On a simple task, interruption improves speed without harming quality: 90.8 seconds against 110.3. The brain remobilises, and the result holds.

On a complex task, the picture inverts. Decision accuracy falls from 0.13 to 0.08, and decision time rises from 608 to 761 seconds. Slower and less accurate. The effect follows the dose: at four interruptions accuracy still holds at 0.22; at twelve it collapses to 0.05.

Translated for a leader: a fragmented day does not degrade the handling of routine subjects. It degrades exactly the category of decisions you are paid for.

The price is paid in stress, not in minutes

A second study completes the picture, and it is counter-intuitive.

In 2008, Gloria Mark, Daniela Gudith and Ulrich Klocke presented at the CHI conference an experiment with 48 participants. The result: interrupted work is finished faster than uninterrupted work, roughly 20.4 minutes against 22.8, at identical quality. Participants compensate for the interruption by speeding up.

But the cost appears elsewhere. Measured stress rises from 6.92 to 9.13, frustration from 4.73 to 6.48, effort from 9.50 to 11.04. All those differences are statistically significant.

In other words, you absorb the acceleration. You pay for it in load, not in delay. And since delay is the only thing your dashboard measures, you never see it coming.

What stays hooked when you change subject

One piece is missing to understand the mechanism. It was named in 2009 by Sophie Leroy, in a paper in Organizational Behavior and Human Decision Processes: attention residue.

When you move from one subject to another, part of your attention stays hooked to the previous one. And, contrary to intuition, finishing the task is not always enough to release that attention: what allows you to let go is the sense of having concluded it, which time pressure paradoxically encourages.

So the leader who runs through eight subjects in a morning never approaches the ninth with all their faculties. They approach it with what is left.

A clarification is owed here, because it is revealing. The most repeated figure in the world on this subject, “it takes 23 minutes and 15 seconds to refocus after an interruption”, has no published primary source. It appears in none of Gloria Mark’s scientific papers, to whom it is universally attributed; it circulates from interviews, and the researcher herself tends to quote an order of magnitude of twenty-five minutes. A journalist wrote it, a thousand articles copied it.

That is a good illustration of this article’s subject: in an accelerated environment, a figure that looks right travels faster than a figure that has been checked.

Three illusions of clarity

From these mechanisms come three reading errors I see constantly.

“I decide as much, so I am fine.” Volume is the indicator that holds longest and informs least. What degrades is the proportion of decisions that hold over time.

“I can feel it when I tire.” Self-assessment is precisely what goes wrong first. An experiment published by METR in July 2025 on experienced developers gives a striking illustration in another field: they expected a 24 % gain from a tool, they were in fact 19 % slower, and afterwards they still estimated they had been 20 % faster. The gap between real and perceived performance is the hardest thing to perceive in oneself.

“I recover at the weekend.” Recovery restores energy, not necessarily the function affected. A study in Sleep in 2015 shows that sleep deprivation specifically impairs the capacity to revise a decision when feedback from the field contradicts the original plan, while working memory stays intact. You keep reasoning correctly. You lose the capacity to change your mind.

A nuance is required, because this ground is crowded with exaggeration. A systematic review published in 2025 in Psychonomic Bulletin & Review, covering 25 studies, concludes that 59.6 % of results show no significant effect of sleep loss on risk-taking. The idea that a tired leader mechanically becomes reckless is not established. What is better established is the impairment of course correction.

The real state of French company leaders

These are not abstract considerations.

The 11th barometer from the Fondation MMA des Entrepreneurs du Futur and Ifop, covering 1,000 leaders of small and mid-sized companies between February and March 2026, gives the following figures: 51 % say they have faced or are facing psychological distress, 24 % of them currently. 85 % report at least one physical or psychological disorder, up 26 points since 2021. Exhaustion affects 50 % of them, insomnia 49 %, anxiety disorders 43 %.

One figure deserves isolating: among leaders reporting sleep disorders, 38 % are in poor psychological health, against 24 % on average, and 67 % report work overload against 55 %.

Sleep is not a comfort indicator. It is the first system to give way, and the easiest to measure.

What AI changes, in both directions

One last variable entered the equation recently, and it cuts both ways.

Artificial intelligence reduces the cost of producing a note, an analysis, a scenario. Mechanically, the quantity of information arriving on the leader’s desk rises, and its average quality becomes harder to assess, because a mediocre document is now well written.

But producing information was never the bottleneck. The bottleneck was turning it into a decision. Accelerating the first without treating the second does not buy clarity: it adds subjects to an already saturated bandwidth.

Used the other way, the same technology does exactly the opposite: it absorbs preparation, synthesis, market watching and follow-up, and gives the leader back the time spent informing themselves rather than deciding. The difference between the two uses is not the tool. It is what you decide to hand it.

A weekly protocol, in three points

None of the above is settled by an effort of will. Three simple arrangements, however, hold up well under pressure.

Put heavy decisions at the start of the run. The order of the day is not neutral, and it is the only lever that produces an effect the following week.

Protect two continuous ninety-minute blocks. Not to work more, but to bring the number of interruptions on complex subjects back into the zone where accuracy still holds.

Institute someone to contradict you, with a fixed slot. A partner, a peer, a board member, an adviser: the status matters little. What matters is that the meeting happens even in the weeks when you judge you do not need it. Those are precisely the ones that count.

Clarity is not a character trait, and it cannot be decreed on Monday morning. It is a resource that gets consumed, that has to be protected, and whose particularity is never to warn you when it runs out.